## NULL

Supplemental Table 1 - Descriptive statistics of excluded vs included samples with methylation data available from the Fragile and Families and Child Wellbeing study. Samples excluded according to the flow diagram in Main Figure 1

Characteristic M samples Final analysis group, M samples = 15051 Not in final analysis group, M samples = 1871
Prenatal maternal smoking 1,692
Not exposed 1,205 (80%) 142 (76%)
Smoke exposed 300 (20%) 45 (24%)
Age at DNA methylation sample 1,692
Age 15 749 (50%) 110 (59%)
Age 9 756 (50%) 77 (41%)
Polymethylation score: coefficients for sustained smoking from newborn cordblood meta-analysis, w/ cell-type control (mean-centered) 1,692 0.00 (0.23) -0.01 (0.24)
AHHR: cg05575921 1,692 77.1 (5.8) 75.3 (7.5)
Pediatric clock 1,692 10.31 (1.93) 10.51 (1.85)
Global methylation 1,692 47.51 (0.80) 47.42 (0.90)
Immune cell proportion (saliva) 1,692 0.94 (0.13) 0.93 (0.15)
Epithelial cell proportion (saliva) 1,692 0.06 (0.13) 0.07 (0.15)
Ancestry categorization from child principal components of genetic data 1,669
African ancestry 907 (60%) 78 (48%)
European ancestry 236 (16%) 19 (12%)
Hispanic ancestry 362 (24%) 67 (41%)
Unknown 0 23
Child gender 1,692
Boy 741 (49%) 107 (57%)
Girl 764 (51%) 80 (43%)
Maternal prenatal alcohol use 1,688 173 (11%) 20 (11%)
Unknown 0 4
Maternal prenatal any drug use 1,686 73 (4.9%) 11 (6.1%)
Unknown 0 6
Constructed - Poverty ratio - mother's household income/poverty threshold 1,692 2.22 (2.43) 2.35 (2.55)
Any postnatal maternal smoking when child age 1 or 5 1,604
Maternal smoking at age 1 or age 5 592 (39%) 52 (53%)
No maternal smoking at age 1 and 5 913 (61%) 47 (47%)
Unknown 0 88
Maternal/primary care giver smoking in month prior to visit (pks/day) 1,674
Less than pack a day 338 (22%) 43 (25%)
No smoking 1,059 (70%) 115 (68%)
Pack or more a day 108 (7.2%) 11 (6.5%)
Unknown 0 18

1 n (%); Mean (SD)

Supplemental Figure 1 - Pearson correlation of selected DNA methylation summary measures from an anlysis of M=1505 samples and N=811children in the Fragile Families and Child Wellbeing cohort. Lower diaganol shows the Pearson rho correlation coefficient, upper triangle shows ellipses illustrating relationship between variables colored by the magnitude and direction of the correlation.

## png 
##   2

Supplemental Figure 2 - Correlation between age 9 and age 15 visit DNA methylation summary measures among 694 children with age 9 and age 15 visit data in the Fragile Families and Child Wellbeing study. Each point represents one individual, x-axis is the DNA methylation summary measure at age 9 and y-axis is the DNA methylation measure at age 15. Blue lines are simple linear regression lines of best fit, with Pearson rho and associated \(P\) value shown. From top-left: Polymethylation score for prenatal maternal smoking, score created using coefficients from a regression of prenatal maternal smoking and DNA methylation in newborn cord blood, incorporating cell-type control. Polymethylation score for prenatal maternal smoking using coefficients from a regression of prenatal maternal smoking and DNA methylation in newborn cord blood, no cell-type control. Polymethylation score calculated using regression coefficients from a regression of prenatal maternal smoking and DNA methylation in older children’s peripheral blood samples. Pediatric epigenetic clock (years). GRIM epigenetic clock (years). Percent DNA methylation at cg05575921 in AHRR. Percent global DNA methylation (mean across all DNA methylation sites in cleaned probe set). Proportion of epithelial cells calculated from DNA methylation using reference panel of children’s saliva. Proportion of immune cells calculated from DNA methylation using a reference panel of children’s saliva.

Supplemental Figure 3: Distributions of selected methylation summary measures at age 9 and age 15 among 811 children in the Fragile Families and Child Wellbeing study. Captive box plots within violin plots show the distribution of selected DNA methylation measures in samples from age 9 (red) and age 15 (blue). Black lines show lines of best fit. From top-left: Polymethylation score for prenatal maternal smoking, score created using coefficients from a regression of prenatal maternal smoking and DNA methylation in newborn cord blood, incorporating cell-type control. Polymethylation score for prenatal maternal smoking using coefficients from a regression of prenatal maternal smoking and DNA methylation in newborn cord blood, no cell-type control. Polymethylation score calculated using regression coefficients from a regression of prenatal maternal smoking and DNA methylation in older children’s peripheral blood samples. Pediatric epigenetic clock (years). GRIM epigenetic clock (years). Percent DNA methylation at cg05575921 in AHRR. Percent global DNA methylation (mean across all DNA methylation sites in cleaned probe set). Proportion of epithelial cells calculated from DNA methylation using reference panel of children’s saliva. Proportion of immune cells calculated from DNA methylation using a reference panel of children’s saliva.

Characteristic Age 9 Age 15
N Not exposed, N = 6001 Smoke exposed, N = 1561 p-value2 N Not exposed, N = 6051 Smoke exposed, N = 1441 p-value2
Child characteristics
Polymethylation score for smoke exposure 756 -0.04 (0.22) 0.09 (0.23) <0.001 749 -0.01 (0.22) 0.11 (0.23) <0.001
AHRR gene: percent cg05575921 methylation 756 77.8 (5.2) 76.8 (5.6) 0.039 749 77 (6) 76 (6) 0.2
Pediatric epigenetic clock (years) 756 9.00 (0.85) 9.01 (1.22) 0.9 749 11.63 (1.77) 11.64 (1.80) >0.9
Percent global DNA methylation 756 47.51 (0.74) 47.59 (0.78) 0.2 749 47.48 (0.85) 47.58 (0.83) 0.2
Immune cell proportion (saliva) 756 0.96 (0.10) 0.96 (0.08) 0.3 749 0.92 (0.15) 0.93 (0.14) >0.9
Epithelial cell proportion (saliva) 756 0.04 (0.10) 0.04 (0.08) 0.3 749 0.08 (0.15) 0.07 (0.14) >0.9
Ancestry categorization from child principal components of genetic data 756 <0.001 749 0.002
African ancestry 356 (59%) 98 (63%) 364 (60%) 89 (62%)
European ancestry 82 (14%) 35 (22%) 85 (14%) 34 (24%)
Hispanic ancestry 162 (27%) 23 (15%) 156 (26%) 21 (15%)
Child gender 756 0.7 749 0.9
Boy 301 (50%) 76 (49%) 295 (49%) 69 (48%)
Girl 299 (50%) 80 (51%) 310 (51%) 75 (52%)
Maternal characteristics
Maternal prenatal alcohol use 756 44 (7.3%) 47 (30%) <0.001 749 40 (6.6%) 42 (29%) <0.001
Maternal prenatal any drug use 756 10 (1.7%) 26 (17%) <0.001 749 9 (1.5%) 28 (19%) <0.001
Maternal income to poverty threshold (at baseline) 756 2.40 (2.57) 1.52 (1.56) <0.001 749 2.41 (2.58) 1.48 (1.57) <0.001
Postnatal maternal smoking at ages 1 or 5 756 <0.001 749 <0.001
No maternal smoking at age 1 and 5 449 (75%) 7 (4.5%) 450 (74%) 7 (4.9%)
Maternal smoking at age 1 or age 5 151 (25%) 149 (96%) 155 (26%) 137 (95%)
Postnatal primary caregiver smoking in past month prior to visit 756 <0.001 749 <0.001
No smoking 478 (80%) 23 (15%) 513 (85%) 45 (31%)
Less than pack a day 95 (16%) 86 (55%) 80 (13%) 77 (53%)
Pack or more a day 27 (4.5%) 47 (30%) 12 (2.0%) 22 (15%)

1 Mean (SD); n (%)

2 Welch Two Sample t-test; Pearson's Chi-squared test

Supplemental Table 2: Bivariate associations between prenatal maternal smoking and additional DNA methylation summary measures among a diverse sample of 811 children in the Fragile Families and Child Wellbeing study. Percent DNA methylation shown calculated across all DNA methylation sites available (global DNA methylation) and stratified by genomic region (Island, Open Sea, Shelf, Shore). Polymethylation score for smoke exposure from main analysis and from sensitivity analyses where scores were calculated using alternative available regression coefficients.

Characteristic Age 9 Age 15
N Not exposed, N = 6001 Smoke exposed, N = 1561 p-value2 N Not exposed, N = 6051 Smoke exposed, N = 1441 p-value2
Percent global DNA methylation 756 47.51 (0.74) 47.59 (0.78) 0.2 749 47.48 (0.85) 47.58 (0.83) 0.2
Island methylation 756 18.96 (0.61) 19.07 (0.59) 0.051 749 19.05 (0.61) 19.14 (0.56) 0.093
Open Sea methylation 756 70.36 (1.05) 70.42 (1.10) 0.5 749 70.22 (1.23) 70.34 (1.22) 0.3
Shelf methylation 756 75.33 (1.04) 75.35 (1.10) 0.8 749 75.19 (1.23) 75.29 (1.21) 0.4
Shore methylation 756 44.06 (0.82) 44.17 (0.86) 0.15 749 44.03 (0.94) 44.15 (0.90) 0.2
Pediatric epigenetic clock (years) 756 9.00 (0.85) 9.01 (1.22) 0.9 749 11.63 (1.77) 11.64 (1.80) >0.9
GRIM epigenetic clock (years) 756 24.7 (3.5) 24.5 (3.3) 0.6 749 30.5 (4.0) 30.3 (4.2) 0.6
GRIM pack/yrs component 756 -5.6 (4.8) -4.7 (5.0) 0.041 749 -4.0 (5.3) -3.6 (5.1) 0.4
Epithelial cell proportion (saliva) 756 0.04 (0.10) 0.04 (0.08) 0.3 749 0.08 (0.15) 0.07 (0.14) >0.9
Immune cell proportion (saliva) 756 0.96 (0.10) 0.96 (0.08) 0.3 749 0.92 (0.15) 0.93 (0.14) >0.9
Polymethylation score for smoke exposure 756 -0.04 (0.22) 0.09 (0.23) <0.001 749 -0.01 (0.22) 0.11 (0.23) <0.001
Polymethylation score: coefficients for sustained smoking from newborn cordblood meta-analysis (mean-centered) 756 -0.06 (0.29) 0.05 (0.27) <0.001 749 0.01 (0.31) 0.12 (0.31) <0.001
Polymethylation score: coeffcients for sustained smoking from older children peripheral blood meta-analysis (mean-centered) 756 -0.02 (0.23) 0.09 (0.25) <0.001 749 -0.02 (0.27) 0.07 (0.28) <0.001
Polymethylation score: coefficients for any smoking from newborn cordblood meta-analysis (mean-centered) 756 -0.04 (0.21) 0.04 (0.19) <0.001 749 0.01 (0.22) 0.09 (0.22) <0.001
AHRR gene: percent cg05575921 methylation 756 77.8 (5.2) 76.8 (5.6) 0.039 749 77 (6) 76 (6) 0.2
MYO1G: cg04180046 756 35 (7) 39 (8) <0.001 749 35 (8) 38 (8) <0.001
CYP1A1: cg05549655 756 18.7 (5.1) 20.9 (5.5) <0.001 749 18.3 (5.2) 20.5 (5.1) <0.001
GFI1: cg14179389 756 5.70 (2.86) 4.88 (2.62) <0.001 749 5.74 (2.86) 4.73 (1.78) <0.001
MYO1G: cg22132788 756 69 (10) 73 (9) <0.001 749 67 (11) 70 (12) 0.006

1 Mean (SD)

2 Welch Two Sample t-test

Supplemental Figure 4: Sensitivity analyses for controlling for different variable sets in multivariable analyses of prenatal maternal smoking and DNA methylation summary measures. Base model: Child sex, maternal income to poverty ratio, immune cell proportion, plate, 1st two principal components from PCA on child genetic data, child age. Prenatal exposures model: base model + yes/no any prenatal maternal alcohol use, yes/no any prenatal other drug use. Secondhand smoking models: prenatal exposures model + yes/no any postnatal maternal/primary care giver smoking at age 1 or 5, pk/day in month prior to primary caregiver/maternal interview at age 9 or 15. Surrogate variable: including all surrogate variables

Supplemental Figure 5 DNA methylation summary measures across combinations of exposure to prenatal and postnatal maternal smoking. Sample size is shown at the bottom of violin plots with captured boxplots. POlymethylation score shown in top facets, percent DNA methylation at cg05575921 in AHRR gene shown in bottom facets, age 9 samples shown in left facets, age 15 shown in right facets

Supplemental Figure 6: Heatmap of P values from Pearson correlations between known covariates and surrogate variables from surrogate variable models

## png 
##   2

Supplemental Table 3: Coefficients for multivariable models of prenatal maternal smoke exposure regressed on selected DNA methylation summary variables (outcomes) among children in the Fragile Families and Child Wellbeing study using surrogate variables

Characteristic Global methylation Pediatric clock Sustained smoking polymethylation score (newborns, cell-type controlled) AHRR: cg05575921
N Beta 95% CI1 p-value N Beta 95% CI1 p-value N Beta 95% CI1 p-value N Beta 95% CI1 p-value
African ancestry
Prenatal maternal smoke exposed 905 0.01 0.01, 0.02 <0.001 905 -0.17 -0.36, 0.02 0.084 905 0.40 0.30, 0.50 <0.001 905 -1.1 -1.6, -0.59 <0.001
European ancestry
Prenatal maternal smoke exposed 234 0.01 0.01, 0.02 <0.001 234 0.03 -0.19, 0.24 0.8 234 0.60 0.38, 0.81 <0.001 234 -1.1 -1.9, -0.36 0.004
Hispanic ancestry
Prenatal maternal smoke exposed 362 0.00 -0.01, 0.01 0.4 362 -0.05 -0.38, 0.27 0.7 362 0.36 0.16, 0.55 <0.001 362 -0.56 -1.5, 0.37 0.2
Unstratified
Prenatal maternal smoke exposed 1501 0.01 0.01, 0.02 <0.001 1501 -0.14 -0.27, 0.00 0.056 1501 0.42 0.33, 0.50 <0.001 1501 -0.95 -1.3, -0.58 <0.001

1 CI = Confidence Interval

Supplemental Table 4: Associations between prenatal maternal smoking and DNA methylation summary measures in multivariable linear models stratified by age or in multivariable linear mixed effect models among 811 US children. All models controlled for: first two principal components of child genetic ancestry (recalculated within ancestry strata for stratified models), child sex, maternal income-to-poverty ratio at birth, immune cell proportion estimated from methylation data, child age

Characteristic Global methylation Pediatric clock Sustained smoking polymethylation score (newborns, cell-type controlled) AHRR: cg05575921
N Beta 95% CI1 p-value N Beta 95% CI1 p-value N Beta 95% CI1 p-value N Beta 95% CI1 p-value
Age 9 models
Prenatal maternal smoke exposed 756 0.02 -0.08, 0.12 0.7 756 0.00 -0.15, 0.14 >0.9 756 0.52 0.36, 0.67 <0.001 756 -1.1 -1.7, -0.41 0.001
Age 15 models
Prenatal maternal smoke exposed 749 0.08 -0.03, 0.19 0.14 749 0.00 -0.22, 0.21 >0.9 749 0.46 0.30, 0.62 <0.001 749 -0.84 -1.6, -0.06 0.034
Longitudinal models
Prenatal maternal smoke exposed 1505 0.05 -0.03, 0.14 0.2 1505 0.02 -0.12, 0.17 0.8 1505 0.48 0.34, 0.63 <0.001 1505 -1.0 -1.5, -0.39 0.001

1 CI = Confidence Interval

Supplemental Table 5: Associations between prenatal maternal smoking and DNA methylation summary measures in multivariable models among 811 US children by ancestry using linear mixed effect models. All models controlled for: first two principal components of child genetic ancestry (recalculated within ancestry strata for stratified models), child sex, maternal income-to-poverty ratio at birth, immune cell proportion estimated from methylation data, child age

Characteristic Global methylation Pediatric clock Sustained smoking polymethylation score (newborns, cell-type controlled) AHRR: cg05575921
N Beta 95% CI1 p-value N Beta 95% CI1 p-value N Beta 95% CI1 p-value N Beta 95% CI1 p-value
African ancestry
Prenatal maternal smoke exposed 907 0.05 -0.06, 0.16 0.4 907 0.13 -0.06, 0.32 0.2 907 0.52 0.34, 0.71 <0.001 907 -1.3 -2.0, -0.55 <0.001
European ancestry
Prenatal maternal smoke exposed 236 -0.04 -0.24, 0.16 0.7 236 0.00 -0.24, 0.24 >0.9 236 0.41 0.05, 0.77 0.027 236 -1.2 -2.4, 0.00 0.051
Hispanic ancestry
Prenatal maternal smoke exposed 362 0.11 -0.10, 0.31 0.3 362 -0.33 -0.72, 0.06 0.10 362 0.25 -0.07, 0.58 0.13 362 0.37 -1.0, 1.8 0.6
Unstratified
Prenatal maternal smoke exposed 1505 0.05 -0.03, 0.14 0.2 1505 0.02 -0.12, 0.17 0.8 1505 0.48 0.34, 0.63 <0.001 1505 -1.0 -1.5, -0.39 0.001

1 CI = Confidence Interval

Supplemental Figure 7: Associations between prenatal maternal smoking and polymethylation scores calculated using alternative coefficient sets among n n_total children in the Fragile Families and Child Wellbeing study. Base model: Child sex, maternal income to poverty ratio, immune cell proportion, plate, 1st two principal components from PCA on child genetic data, child age. Prenatal exposures model: base model + yes/no any prenatal maternal alcohol use, yes/no any prenatal other drug use. Secondhand smoking models: prenatal exposures model + yes/no any postnatal maternal/primary care giver smoking at age 1 or 5, pk/day in month prior to primary caregiver/maternal interview at age 9 or 15. Surrogate variable: including all surrogate variables

Supplemental Figure 8 - Associations between prenatal maternal smoking and other a priori CpG sites selected from previous meta-analyses among 811 children in the Fragile Families and Child Wellbeing study

Supplemental Table 6: Area under the curve for models shown in Main Figure 4 for classifying prenatal maternal smoke exposure, with P value when comparing to base model only or to model using polymethylation score + base model. Base model: Child sex, maternal income to poverty ratio, immune cell proportion, plate, 1st two principal components from PCA on child genetic data, child age.

Age Model AUC P value vs base model P value vs base model + polymethylation score
Age 9 Global methylation 0.54 0.0000 0.0000
Age 9 Sustained smoking polymethylation score (newborns, cell-type controlled) 0.67 0.0560 0.0000
Age 9 AHRR: cg05575921 0.55 0.0000 0.0000
Age 9 Base model+No methylation 0.73 1.0000 0.0003
Age 9 Base model+Global methylation 0.73 0.5607 0.0003
Age 9 Base model+Sustained smoking polymethylation score (newborns, cell-type controlled) 0.77 0.0003 1.0000
Age 9 Base model+AHRR: cg05575921 0.74 0.1328 0.0334
Age 15 Global methylation 0.53 0.0000 0.0000
Age 15 Sustained smoking polymethylation score (newborns, cell-type controlled) 0.65 0.0271 0.0000
Age 15 AHRR: cg05575921 0.54 0.0000 0.0000
Age 15 Base model+No methylation 0.72 1.0000 0.0009
Age 15 Base model+Global methylation 0.72 0.3802 0.0010
Age 15 Base model+Sustained smoking polymethylation score (newborns, cell-type controlled) 0.77 0.0009 1.0000
Age 15 Base model+AHRR: cg05575921 0.73 0.4054 0.0071

Supplemental Table 7: Area under the curve when using different coefficient sets to construct polymethylation scores for predicting prenatal maternal smoke exposure. With P value when comparing to base model only or to model using polymethylation score + base model. Base model: Child sex, maternal income to poverty ratio, immune cell proportion, plate, 1st two principal components from PCA on child genetic data, child age.

Age Model AUC P value vs base model P value vs base model + polymethylation score with cell-type control
Age 9 Any smoking polymethylation score (newborns) 0.6215 0.0012 0.0000
Age 9 Sustained smoking polymethylation score (newborns, cell-type controlled) 0.6658 0.0560 0.0000
Age 9 Sustained smoking polymethylation score (newborns) 0.6196 0.0010 0.0000
Age 9 Sustained smoking polymethylation score (older children) 0.6341 0.0040 0.0000
Age 9 Base model+No methylation 0.7252 1.0000 0.0003
Age 9 Base model+Any smoking polymethylation score (newborns) 0.7598 0.0027 0.0142
Age 9 Base model+Sustained smoking polymethylation score (newborns, cell-type controlled) 0.7737 0.0003 1.0000
Age 9 Base model+Sustained smoking polymethylation score (newborns) 0.7596 0.0027 0.0125
Age 9 Base model+Sustained smoking polymethylation score (older children) 0.7666 0.0007 0.4181
Age 15 Any smoking polymethylation score (newborns) 0.6062 0.0007 0.0000
Age 15 Sustained smoking polymethylation score (newborns, cell-type controlled) 0.6473 0.0271 0.0000
Age 15 Sustained smoking polymethylation score (newborns) 0.6056 0.0006 0.0000
Age 15 Sustained smoking polymethylation score (older children) 0.5923 0.0002 0.0000
Age 15 Base model+No methylation 0.7213 1.0000 0.0009
Age 15 Base model+Any smoking polymethylation score (newborns) 0.7566 0.0011 0.1124
Age 15 Base model+Sustained smoking polymethylation score (newborns, cell-type controlled) 0.7664 0.0009 1.0000
Age 15 Base model+Sustained smoking polymethylation score (newborns) 0.7569 0.0011 0.1270
Age 15 Base model+Sustained smoking polymethylation score (older children) 0.7530 0.0075 0.1071